The purpose of this study is to determine which of several image processing and display methods is the most reliable for assessing progression of brain tumors, based on magnetic resonance imaging (MRI). Accurate assessment of MRI studies is crucial to the efficient conduct of clinical trials because it is one of the few metrics that is non-invasive, and that can be easily conducted prior to, during, and after therapy. The seemingly easy task of image evaluation can be difficult because variance in how the images are acquired (e.g. patient position) and complex signal within the lesion can obscure important findings. Experienced observers often disagree about the presence of, and the meaning of, changes seen on MRI. Using an automated algorithm to determine changes may improve the ability to assess response to therapy. Increased sensitivity can allow earlier determination of the effectiveness of new treatments. Stopping ineffective therapies can reduce patient morbidity and reduce costs. An algorithmic method also permits easier comparison across sites and methods, as the variable of how and who assessed the images are removed or reduced. Conventional measurements of tumors, such as RECIST or Macdonald methods, poorly estimate the total amount of tumor tissue (and therefore, changes in tumor volume), particularly for brain tumors that are frequently irregular, particularly after surgery. Computer volume measurement may more accurately measure tumor volumes, but make important assumptions that may not hold, for instance, that all of a tumor enhances, and only tumor enhances. This may account for the poor predictive value of volume measurement of brain tumors. Computer methods have been developed that may be helpful in reducing the variability of raters, and for increasing sensitivity to change. These include methods for automatically aligning old and new studies. If images are aligned, subtraction may further help to elucidate changes. In this study, we also test a method we have developed which is a multi-parametric change detector. We also will study the effects on how images are presented (side-by-side versus in-place or flicker mode). While this study focuses on brain tumors, we believe the results may be applicable to many other tumor types (and in fact, other diseases). It is our hope that in the future, a measurement method superior to or as an adjunct to human observation will become available, though this study will not attempt to go that far. ? ? ?
| Erickson, B J; Wood, C P; Kaufmann, T J et al. (2011) Optimal presentation modes for detecting brain tumor progression. AJNR Am J Neuroradiol 32:1652-7 |
| Bower, R S; Burrus, T M; Giannini, C et al. (2010) Teaching NeuroImages: demyelinating disease mimicking butterfly high-grade glioma. Neurology 75:e4-5 |
